[R-sig-Geo] Creating Regions With a Dummy Variable Identifier

[Roger Bivand]

CONTINUES AT FOOT:

On Fri, 21 Jan 2011, Matthew Hasbrouck wrote:

>> Date: Fri, 21 Jan 2011 08:11:20 +0100
>> From: Roger.Bivand at nhh.no
>> To: mbh02 at live.com
>> CC: r-sig-geo at r-project.org
>> Subject: Re: [R-sig-Geo] Creating Regions With a Dummy Variable Identifier
>>
>> On Thu, 20 Jan 2011, Matthew Hasbrouck wrote:
>>
>>>
>>> Hello everyone,
>>>
>>> I am a graduate student currently working with census tract data. I have
>>> identified census tracts as either having a certain characteristic or
>>> not, using a dummy variable. This variable is coded as 1 if the tract
>>> has that characteristic and 0 if it doesn't. I would like to take all
>>> contiguous tracts coded as 1 and dissolve them into single records. What
>>> would be left are all of the original tracts that do not have this
>>> characteristic, coded as 0, and records for the larger dissolved areas
>>> that contained this characteristic. I originally thought the SKATER
>>> function might work, but I don't think it will work properly with a
>>> dummy variable.
>>
>> Please see ?unionSpatialPolygons in maptools. You'll need to construct a
>> new set of ID values, with your tracts to be dissolved having the same ID,
>> and all the others having their own different IDs. Note the warning about
>> the license of gpclib in case you are affected.
>>
>> Roger
>>
>>>
>>> Thanks,
>>> Matt
>
>
> I think I understand a way to do this using unionSpatialPolygons if I
> have a fairly small number of tracts.  I could view a map of the dummy
> variable and manually give a unique ID to all tracts in a particular
> cluster.  I could continue doing that for all other clusters.  Then I
> could save the shapefile and union them in R.  The problem is
> that I am dealing with over 9,000 tracts in 91 cities.  Is there a
> systematic way to identify all of the contiguous 1's (on the dummy
> variable) and give them an ID?
>

So you have two questions, not one? You want to identify contiguous groups 
of 1s on the dummy, and group them into separate Polygons objects, and 
then merge the m contiguous groups so identified. I suggest subsetting the 
input object to just the 1s, then use poly2nb() to get the contiguities, 
next n.comp.nb() to identify the groups, unionSpatialPolygons() them, and 
spRbind() back the 0s. I would have liked to provide an example but have 
to rush ...

Hope this helps,

Roger



> Thanks,
> Matt
>
>
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no